In this paper, GPS, WLAN, and INS are used as positioning information data sources, and multi-source fusion technology is used to build an intelligent positioning navigation device. The feature vectors of the original data are observed using the federal Kalman filter algorithm, and the information is continuously updated to assign weights to achieve the optimization of the positioning information. The three-dimensional information of the defined position is measured using the CNN algorithm, the faulty data in the filter is detected and removed, and the detected data enters the main filter instead of the faulty data for fusion. To prove the effectiveness of the device, experimental analysis is performed, and the results show that the device provides diverse, intelligent services with up to 95% localization accuracy and stronger signals.